]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
25 years agoSync with sys/i386/conf/userconfig.c revision 1.152.
kato [Tue, 10 Aug 1999 08:35:57 +0000 (08:35 +0000)]
Sync with sys/i386/conf/userconfig.c revision 1.152.

25 years agoCheck to see if /usr/bin/logger exists before using it. Use "echo" if
obrien [Tue, 10 Aug 1999 07:16:00 +0000 (07:16 +0000)]
Check to see if /usr/bin/logger exists before using it.  Use "echo" if
/usr/bin/logger is not available.

Noticed from: error messages in -CURRENT's sysinstall Alt-F2 debug screen

25 years agoUse the latest version of these files from OpenBSD.
imp [Tue, 10 Aug 1999 05:58:58 +0000 (05:58 +0000)]
Use the latest version of these files from OpenBSD.

1) Safty change from casper dik was added to OpenBSD's sources since I
   grabbed them. milltert@openbsd.org
2) Split up strlcpy to improve efficiency of the common case.
   milltert@openbsd.org
3) Cleanup of cross references for man page.  {alex,aaron}@openbsd.org

Pointed out by: deraadt@openbsd.org

25 years agoImport strl{cat,cpy} from OpenBSD.
imp [Tue, 10 Aug 1999 05:21:31 +0000 (05:21 +0000)]
Import strl{cat,cpy} from OpenBSD.

Obtained from: OpenBSD

25 years agovm_map_madvise:
alc [Tue, 10 Aug 1999 04:50:20 +0000 (04:50 +0000)]
vm_map_madvise:
Now that behaviors are stored in the vm_map_entry rather than
the vm_object, it's no longer necessary to instantiate a vm_object
just to hold the behavior.

Reviewed by: dillon

25 years agopmap_remove_pages:
alc [Tue, 10 Aug 1999 04:10:57 +0000 (04:10 +0000)]
pmap_remove_pages:
Add KASSERT to detect out of range access to the pv_table and
report the errant pte before it's overwritten.

25 years agoSome floppy related commands (fdformat, fdwrite, fdcontrol) need not be
mjacob [Tue, 10 Aug 1999 01:04:11 +0000 (01:04 +0000)]
Some floppy related commands (fdformat, fdwrite, fdcontrol) need not be
restricted to i386.

25 years agoAdd support for the Am79C978. (AMD PCHome/PCI Ethernet adapter)
mdodd [Tue, 10 Aug 1999 01:03:51 +0000 (01:03 +0000)]
Add support for the Am79C978. (AMD PCHome/PCI Ethernet adapter)

See: http://www.amd.com/products/npd/overview/homenetworking/intro.html

PR: kern/12275
Submitted by: Robert Watson <robert@cyrus.watson.org>

25 years agoBack out the last patch. I'm too tired to apply patches.
brian [Mon, 9 Aug 1999 23:49:59 +0000 (23:49 +0000)]
Back out the last patch.  I'm too tired to apply patches.

25 years agoRemove CLOCAL when we're not in -dedicated mode.
brian [Mon, 9 Aug 1999 23:38:56 +0000 (23:38 +0000)]
Remove CLOCAL when we're not in -dedicated mode.

PR: 13046
Submitted by: Vsevolod Lobko <seva@alex-ua.com>

25 years agoCorrect typo.
grog [Mon, 9 Aug 1999 23:28:03 +0000 (23:28 +0000)]
Correct typo.

PR: docs/13050
Reported-by: Dan Langille<junkmale@xtra.co.nz>

25 years agoAdd an example of an isdnd.rc that works with ppp.
brian [Mon, 9 Aug 1999 22:58:55 +0000 (22:58 +0000)]
Add an example of an isdnd.rc that works with ppp.

25 years agoAdd an example of an isdn profile.
brian [Mon, 9 Aug 1999 22:57:47 +0000 (22:57 +0000)]
Add an example of an isdn profile.

25 years agoChange printf formats %q[du] -> %ll[du]
brian [Mon, 9 Aug 1999 22:56:18 +0000 (22:56 +0000)]
Change printf formats %q[du] -> %ll[du]

25 years agoInsist that ppp.conf and it's parent directories aren't `other'
brian [Mon, 9 Aug 1999 22:54:51 +0000 (22:54 +0000)]
Insist that ppp.conf and it's parent directories aren't `other'
writable rather than not being writable at all.

PR: 13009
Kind-of submitted by: Anthony Kimball <alk@pobox.com>

25 years agoFix nexus_pcib_is_host_bridge() so that it detects my 486's PCI bus
wpaul [Mon, 9 Aug 1999 21:12:18 +0000 (21:12 +0000)]
Fix nexus_pcib_is_host_bridge() so that it detects my 486's PCI bus
correctly. It has the following code:

        if (class != PCIC_BRIDGE || subclass != PCIS_BRIDGE_HOST)
                return NULL;

My 486 has an Integrated Micro Solutions PCI bridge which identifies
itself as subclass PCIS_BRIDGE_OTHER, not PCIS_BRIDGE_HOST. Consequently,
it gets ignored. In my opinion, the correct test should be:

        if ((class != PCIC_BRIDGE) && (subclass != PCIS_BRIDGE_HOST))
                return NULL;

That way the test still succeeds because the chip's class is PCIC_BRIDGE.
Clearly it's not reasonable to expect all host to PCI bridges to always
have a subclass of PCIS_BRIDGE_HOST since I've got one that doesn't.
This way the sanity test should remain relatively sane while still allowing
some oddball yet correct hardware to work. If somebody has a better way
to do it, go ahead and tweak the test, but be aware that
class == PCIC_BRIDGE and subclass == PCIS_BRIDGE_OTHER is a valid case.

While I was here, I also added an explicit ID string for the IMS chipset.
I also dealt with a minor style nit: it's bad karma not to have a default
case for your switch statements, but the one in this routine doesn't have
one. The default string of "Host to PCI bridge" is now assigned in a
default case of the switch statement instead of initializing "s" with the
string before the switch and then not having any default case.

25 years agochip0: <PCI to 0x80 bridge (vendor=10e0 device=8849)> at device 0.0 on pci0
mdodd [Mon, 9 Aug 1999 21:11:44 +0000 (21:11 +0000)]
chip0: <PCI to 0x80 bridge (vendor=10e0 device=8849)> at device 0.0 on pci0

Isn't really that useful.

chip0: <PCI to Other bridge (vendor=10e0 device=8849)> at device 0.0 on pci0

is more in keeping with the spirit of the rest of the code.

Previous behavior with regard to truely unknown bridges unchanged.

"<Anti-Bill> Tell you what: you have commit privs now. You do it."

25 years agoMove cronjob up one minute to get it out of the 02:00-03:00 happy hour
phk [Mon, 9 Aug 1999 19:56:29 +0000 (19:56 +0000)]
Move cronjob up one minute to get it out of the 02:00-03:00 happy hour
of daylight savings changes.

Submitted by: Doug <Doug@gorean.org>
Reviewed by: phk

25 years agoInclude <sys/queue.h> since this header now depends on it.
jdp [Mon, 9 Aug 1999 18:45:20 +0000 (18:45 +0000)]
Include <sys/queue.h> since this header now depends on it.

25 years agoThere is no special handling for SIGALRM as
ru [Mon, 9 Aug 1999 17:31:30 +0000 (17:31 +0000)]
There is no special handling for SIGALRM as
sleep(3) is implemented using nanosleep(2).

25 years agoA little more tidying up.
peter [Mon, 9 Aug 1999 14:43:39 +0000 (14:43 +0000)]
A little more tidying up.

25 years agoFix a few typos and misspellings, grabbed from PR#8472
chris [Mon, 9 Aug 1999 14:31:04 +0000 (14:31 +0000)]
Fix a few typos and misspellings, grabbed from PR#8472

25 years agoExpand a heap of macros that obscure readability and are no longer needed
peter [Mon, 9 Aug 1999 14:15:46 +0000 (14:15 +0000)]
Expand a heap of macros that obscure readability and are no longer needed
here, and do a bit of general tidy up.

25 years agoRemove a reference to config(8) when IRQ 2 is remapped to IRQ 9.
mpp [Mon, 9 Aug 1999 13:52:12 +0000 (13:52 +0000)]
Remove a reference to config(8) when IRQ 2 is remapped to IRQ 9.
Config(8) contains no documentation about this.

Fix the help for the PnP irq and drq commands.  This one caused
me a bit of head scratching the other night while trying to get
a problematic PnP device configured properly.

25 years agoMore pre-lite2 support zapped and some more tidy-up.
peter [Mon, 9 Aug 1999 13:24:55 +0000 (13:24 +0000)]
More pre-lite2 support zapped and some more tidy-up.

25 years agoUnifdef -D__FreeBSD__ - and remove pre-lite2 support.
peter [Mon, 9 Aug 1999 13:15:53 +0000 (13:15 +0000)]
Unifdef -D__FreeBSD__ - and remove pre-lite2 support.

25 years agoUnifdef -U__NetBSD__
peter [Mon, 9 Aug 1999 13:07:37 +0000 (13:07 +0000)]
Unifdef -U__NetBSD__

25 years agoFixed checking a type of the interface.
nyan [Mon, 9 Aug 1999 13:03:35 +0000 (13:03 +0000)]
Fixed checking a type of the interface.

Submitted by:   Tomohiko Kurahashi <kura@tim.hi-ho.ne.jp>

25 years agoUnifdef -U__bsdi__
peter [Mon, 9 Aug 1999 13:01:58 +0000 (13:01 +0000)]
Unifdef -U__bsdi__

25 years agos/Id/FreeBSD/
peter [Mon, 9 Aug 1999 12:52:49 +0000 (12:52 +0000)]
s/Id/FreeBSD/

25 years agoStrip __FreeBSD_version >= 300000 conditionals.
peter [Mon, 9 Aug 1999 12:50:12 +0000 (12:50 +0000)]
Strip __FreeBSD_version >= 300000 conditionals.

25 years agoMerge changes from NetBSD rev 1.82 -> 1.86 via vendor branch.
peter [Mon, 9 Aug 1999 12:29:29 +0000 (12:29 +0000)]
Merge changes from NetBSD rev 1.82 -> 1.86 via vendor branch.
Among the changes:  1.84: support compex 4-port cards.

25 years agomake alpha compile again.
phk [Mon, 9 Aug 1999 11:02:45 +0000 (11:02 +0000)]
make alpha compile again.

25 years agoMerge the cons.c and cons.h to the best of my ability. alpha may or
phk [Mon, 9 Aug 1999 10:35:05 +0000 (10:35 +0000)]
Merge the cons.c and cons.h to the best of my ability.  alpha may or
may not compile, I can't test it.

25 years agoBack out previous commit - it's not necessary now that tty.h properly includes
jkh [Mon, 9 Aug 1999 07:51:03 +0000 (07:51 +0000)]
Back out previous commit - it's not necessary now that tty.h properly includes
the queue macros.

25 years agoThis uses an SLIST_ENTRY and breaks the build without sys/queue.h. Commit
jkh [Mon, 9 Aug 1999 06:50:07 +0000 (06:50 +0000)]
This uses an SLIST_ENTRY and breaks the build without sys/queue.h.  Commit
this until bde and friends finish arguing over what to do instead. :)

25 years agoRemove reference to nonexistant man page `da(9)'
chris [Mon, 9 Aug 1999 02:35:55 +0000 (02:35 +0000)]
Remove reference to nonexistant man page `da(9)'

25 years agoAdd a missing include (sys/queue.h for sys/tty.h SLIST usage.)
green [Mon, 9 Aug 1999 02:19:38 +0000 (02:19 +0000)]
Add a missing include (sys/queue.h for sys/tty.h SLIST usage.)

25 years agoRemove f2c-related stuff from the "includes" target and from a
jdp [Sun, 8 Aug 1999 22:26:09 +0000 (22:26 +0000)]
Remove f2c-related stuff from the "includes" target and from a
comment.  Fixes broken make world.

25 years agoRemove compile time limit on number of devices.
phk [Sun, 8 Aug 1999 22:25:46 +0000 (22:25 +0000)]
Remove compile time limit on number of devices.

25 years agoPassing the unit number to an interrupt routing is kind of narrowminded,
phk [Sun, 8 Aug 1999 22:25:21 +0000 (22:25 +0000)]
Passing the unit number to an interrupt routing is kind of narrowminded,
add a way to get the entire struct ppb_device back.

25 years agoRemove compiled in limit on number of devices.
phk [Sun, 8 Aug 1999 22:01:50 +0000 (22:01 +0000)]
Remove compiled in limit on number of devices.

25 years agoregister tty devices so they will look right in pstat.
phk [Sun, 8 Aug 1999 21:35:17 +0000 (21:35 +0000)]
register tty devices so they will look right in pstat.

25 years agoDocument -f flag:
chris [Sun, 8 Aug 1999 21:08:05 +0000 (21:08 +0000)]
Document -f flag:

     -f      Do not display a diagnostic message if chmod could not modify the
             mode for file.

25 years agoAllocate and register struct tty on the fly.
phk [Sun, 8 Aug 1999 20:25:14 +0000 (20:25 +0000)]
Allocate and register struct tty on the fly.

25 years agoEnable ttymalloc().
phk [Sun, 8 Aug 1999 20:24:58 +0000 (20:24 +0000)]
Enable ttymalloc().

25 years agos/TULUP/TULIP/ - from vendor branch.
peter [Sun, 8 Aug 1999 19:56:06 +0000 (19:56 +0000)]
s/TULUP/TULIP/ - from vendor branch.

25 years agoAdd new sysctl "kern.ttys" which return all the struct tty's which have
phk [Sun, 8 Aug 1999 19:47:32 +0000 (19:47 +0000)]
Add new sysctl "kern.ttys" which return all the struct tty's which have
been registered with ttyregister().

register ptys with ttyregister().

25 years agoAdd support for picking up ttys with sysctl(kern.ttys).
phk [Sun, 8 Aug 1999 19:44:00 +0000 (19:44 +0000)]
Add support for picking up ttys with sysctl(kern.ttys).

25 years agoMake the pty driver as close to a cloning device as we can get for now,
phk [Sun, 8 Aug 1999 19:28:59 +0000 (19:28 +0000)]
Make the pty driver as close to a cloning device as we can get for now,
we create the pty on the fly when it is first opened.

If you run out of ptys now, just MAKEDEV some more.

This also demonstrate the use of dev_t->si_tty_tty and dev_t->si_drv1
in a device driver.

25 years agoDecommision miscfs/specfs/specdev.h. Most of it goes into <sys/conf.h>,
phk [Sun, 8 Aug 1999 18:43:05 +0000 (18:43 +0000)]
Decommision miscfs/specfs/specdev.h.  Most of it goes into <sys/conf.h>,
a few lines into <sys/vnode.h>.

Add a few fields to struct specinfo, paving the way for the fun part.

25 years agoAvoid symbol clash in LINT.
phk [Sun, 8 Aug 1999 18:17:40 +0000 (18:17 +0000)]
Avoid symbol clash in LINT.

25 years agoSilence a warning about uninitialized variables.
brian [Sun, 8 Aug 1999 17:36:10 +0000 (17:36 +0000)]
Silence a warning about uninitialized variables.

25 years agoDon't includei 4b support on the alpha
brian [Sun, 8 Aug 1999 17:35:35 +0000 (17:35 +0000)]
Don't includei 4b support on the alpha

25 years agoRemove f2c as its libraries were tossed last April, and f2c has been replaced
obrien [Sun, 8 Aug 1999 17:08:30 +0000 (17:08 +0000)]
Remove f2c as its libraries were tossed last April, and f2c has been replaced
with EGCS's f77.

Noticed still alive by: bde

25 years agoFix a reference to `st' by replacing it with `sa'.
chris [Sun, 8 Aug 1999 16:27:17 +0000 (16:27 +0000)]
Fix a reference to `st' by replacing it with `sa'.

25 years agoDon't log_Printf(..., buf) in case buf contains %s
brian [Sun, 8 Aug 1999 15:19:00 +0000 (15:19 +0000)]
Don't log_Printf(..., buf) in case buf contains %s

Info provided by: Ollivier Robert <roberto@keltia.freenix.fr>

25 years agoFixed all printf format errors reported by gcc -Wformat on i386's:
bde [Sun, 8 Aug 1999 14:11:03 +0000 (14:11 +0000)]
Fixed all printf format errors reported by gcc -Wformat on i386's:
- %q -> %ll.

Fixed nearby errors not reported by gcc -Wformat on i386's:
- don't assume that the promotion of [u_]int64_t is [u_]quad_t.
- don't use signed formats for unsigned args.

25 years agoFixed printf format errors (%qu -> %llu; the arg was already unsigned long
bde [Sun, 8 Aug 1999 13:43:51 +0000 (13:43 +0000)]
Fixed printf format errors (%qu -> %llu; the arg was already unsigned long
long to hide problems on alphas).

25 years agoFixed all printf format errors reported by gcc -Wformat on i386's:
bde [Sun, 8 Aug 1999 13:22:05 +0000 (13:22 +0000)]
Fixed all printf format errors reported by gcc -Wformat on i386's:
- %q -> %ll; don't assume that the promotion of off_t is quad_t; only
  assume that off_t's are representable as long longs.
- printing of dev_t's was completely broken.

Fixed nearby printf format errors not reported by gcc -Wformat on i386's:
- printing of ino_t's and pointers was sloppy.

25 years agoFix page fault in linux_uselib syscall.
marcel [Sun, 8 Aug 1999 11:26:46 +0000 (11:26 +0000)]
Fix page fault in linux_uselib syscall.

PR: 12910
Submitted by: Peter Holm <peter@holm.cc>

25 years agoSync with sys/i386/conf/options.i386 revision 1.119.
kato [Sun, 8 Aug 1999 11:13:47 +0000 (11:13 +0000)]
Sync with sys/i386/conf/options.i386 revision 1.119.

25 years agoSync with sys/i386/conf/majors.i386 revision 1.81.
kato [Sun, 8 Aug 1999 11:12:51 +0000 (11:12 +0000)]
Sync with sys/i386/conf/majors.i386 revision 1.81.

25 years agoSync with sys/i386/conf/files.i386 revision 1.257.
kato [Sun, 8 Aug 1999 11:12:00 +0000 (11:12 +0000)]
Sync with sys/i386/conf/files.i386 revision 1.257.

25 years agoEnable bpf by default.
kato [Sun, 8 Aug 1999 11:09:28 +0000 (11:09 +0000)]
Enable bpf by default.

25 years agoFixed missing "G" in the list item for the main description of %g and
bde [Sun, 8 Aug 1999 11:00:01 +0000 (11:00 +0000)]
Fixed missing "G" in the list item for the main description of %g and
%G formats.

25 years agoFix a panic caused by freeing unallocated structure.
kato [Sun, 8 Aug 1999 10:06:27 +0000 (10:06 +0000)]
Fix a panic caused by freeing unallocated structure.

Submitted by: Tomohiko Kurahashi <kura@tim.hi-ho.ne.jp>

25 years agoAdd vinumraid5.c to the files for Vinum. This allows (deprecated)
grog [Sun, 8 Aug 1999 08:29:58 +0000 (08:29 +0000)]
Add vinumraid5.c to the files for Vinum.  This allows (deprecated)
kernel builds with Vinum.

Linted-by: phk

25 years agoThe following simplifies newvers.sh somewhat and makes what(1) work with
jkh [Sun, 8 Aug 1999 07:51:16 +0000 (07:51 +0000)]
The following simplifies newvers.sh somewhat and makes what(1) work with
kernels again, now that we're using EGCS/GCC 2.9+. This "here"
file is compatible with the Bourne shell and the Korn shell (incl. pdksh
and KSH93 from AT&T, which I do have), so it doesn't make newvers.sh
unportable, but makes it easier to modify in the future/read now.

Submitted by: green

25 years agoForgotten in previous log message:
bde [Sun, 8 Aug 1999 07:31:47 +0000 (07:31 +0000)]
Forgotten in previous log message:

PR: misc/12920

25 years agoInstall c++filt in /usr/libexec/${OBJFORMAT}. The version that
bde [Sun, 8 Aug 1999 07:30:02 +0000 (07:30 +0000)]
Install c++filt in /usr/libexec/${OBJFORMAT}.  The version that
was installed in /usr/bin normally got clobbered when objformat
was installed.  Indirection through objformat is correct although
underscore handling is the only thing that differs for aout and
elf -- going through objformat is the easiest way to set c++filt's
underscore handling flag correctly.

25 years agoCorrection to previous commit.
stark [Sun, 8 Aug 1999 07:23:12 +0000 (07:23 +0000)]
Correction to previous commit.

25 years agoCorrected mistake that was causing daemon to loop without serving
stark [Sun, 8 Aug 1999 07:05:46 +0000 (07:05 +0000)]
Corrected mistake that was causing daemon to loop without serving
user requests.  Note that nothing can be said about the value of 'user'
unless User != NULL.

25 years agocdevsw_remove: place correct value in bmaj2cmaj. This had caused
grog [Sun, 8 Aug 1999 00:34:00 +0000 (00:34 +0000)]
cdevsw_remove: place correct value in bmaj2cmaj.  This had caused
warnings of the following nature on reloading a kld:

  WARNING: "vinum" is usurping "console"'s bmaj

This only applies to cases where "console" is mentioned.

Broken-by:   grog

25 years agoMake long longs ("%ll" format) work.
green [Sat, 7 Aug 1999 20:13:32 +0000 (20:13 +0000)]
Make long longs ("%ll" format) work.

Reviewed by: msmith

25 years agoFixed a bogus include pathname which broke compiling in an obj directory.
bde [Sat, 7 Aug 1999 13:11:12 +0000 (13:11 +0000)]
Fixed a bogus include pathname which broke compiling in an obj directory.

25 years agoFixed syntax errors. Code should be tested before committing it, or at
bde [Sat, 7 Aug 1999 13:09:30 +0000 (13:09 +0000)]
Fixed syntax errors.  Code should be tested before committing it, or at
least before enabling it in LINT and merging it into RELENG_3.

25 years agoRe-delete the (meaningless) device-driver tokens that came back in 1.256
peter [Sat, 7 Aug 1999 12:19:41 +0000 (12:19 +0000)]
Re-delete the (meaningless) device-driver tokens that came back in 1.256
after having been removed in 1.253 and turned into a warning.

Noticed by: bde

25 years agoupdate list of ISDN hardware supported by isdn4bsd 0.83
hm [Sat, 7 Aug 1999 12:14:39 +0000 (12:14 +0000)]
update list of ISDN hardware supported by isdn4bsd 0.83

25 years agoUpdate to reflect recent changes.
grog [Sat, 7 Aug 1999 09:27:25 +0000 (09:27 +0000)]
Update to reflect recent changes.

25 years agoAdd vinumraid5.c to list of sources.
grog [Sat, 7 Aug 1999 08:35:54 +0000 (08:35 +0000)]
Add vinumraid5.c to list of sources.

25 years agoImport RAID-5 code.
grog [Sat, 7 Aug 1999 08:22:49 +0000 (08:22 +0000)]
Import RAID-5 code.
Add Cybernet copyright.

OK'd-by:     Chuck Jacobus <chuck@cybernet.com>

25 years agoRemove code to detect RAID-5/non-RAID-5 kernel modules.
grog [Sat, 7 Aug 1999 08:20:07 +0000 (08:20 +0000)]
Remove code to detect RAID-5/non-RAID-5 kernel modules.

25 years agoRemove code to detect RAID-5/non-RAID-5 kernel modules.
grog [Sat, 7 Aug 1999 08:19:45 +0000 (08:19 +0000)]
Remove code to detect RAID-5/non-RAID-5 kernel modules.

Add function definition for readpol command.

Rewrite make_devices with a view to incremental making.  In the
process, effectively kill off plex-bound subdisk device numbers.

25 years agoImport RAID-5 code.
grog [Sat, 7 Aug 1999 08:19:15 +0000 (08:19 +0000)]
Import RAID-5 code.
Add Cybernet copyright.

OK'd-by:     Chuck Jacobus <chuck@cybernet.com>

vinum_ldi:
  Calculate the percentage used correctly.

vinum_lsi:
  Flag detached subdisks.

vinum_info:
  Request info: get the major and minor numbers and print them
  correctly.

list_defective_objects:
  New function, for listing objects which aren't up after vinum_start.

Requested-by: Steve Taylor <staylor@cybernet.com>
25 years agovinum_init:
grog [Sat, 7 Aug 1999 08:18:39 +0000 (08:18 +0000)]
vinum_init:
   wait correctly for completion.
   set object in initializing state before initializing.
   Don't set the plex up, it should go up automatically.

vinum_start:
  If any objects are not up on start, list them.

Requested-by: Steve Taylor <staylor@cybernet.com>
vinum_attach:
  Make sure we always have the object information before examining it.

vinum_replace:
  Start writing.

create_drive, vinum_concat, vinum_stripe, vinum_mirror:
  Check the ioctl return values correctly

vinum_readpol:
  Start writing.

25 years agoImport RAID-5 code.
grog [Sat, 7 Aug 1999 08:15:25 +0000 (08:15 +0000)]
Import RAID-5 code.
Add Cybernet copyright.

OK'd-by:     Chuck Jacobus <chuck@cybernet.com>

Define VINUM_BLOCK_SD and VINUM_CHAR_SD macros to create raw subdisk
device numbers.

Add VF_HOTSPARE flag.

25 years agoUpdate a comment.
grog [Sat, 7 Aug 1999 08:14:44 +0000 (08:14 +0000)]
Update a comment.

25 years agoImport RAID-5 code.
grog [Sat, 7 Aug 1999 08:14:01 +0000 (08:14 +0000)]
Import RAID-5 code.
Add Cybernet copyright.

OK'd-by:     Chuck Jacobus <chuck@cybernet.com>

update_plex_state:
  If any subdisk in the plex is initializing, set the plex to
  initializing state.  This gets rid of the ugly corrupt/degraded/up
  transitions which previously occurred.

Desired-by: Steve Taylor <staylor@cybernet.com>
sddownstate:
  Add new function, used by checksdstate.

checksdstate:
  Let sddownstate decide what status to return.

25 years agoImport RAID-5 code.
grog [Sat, 7 Aug 1999 08:13:23 +0000 (08:13 +0000)]
Import RAID-5 code.
Add Cybernet copyright.

OK'd-by:     Chuck Jacobus <chuck@cybernet.com>

logrq: save device major and minor numbers to compensate for lost
  dev_t.

launch_requests: Don't issue requests which are marked
  XFR_BAD_SUBDISK.  This may make things easier in bre().

bre:
  Rearrange.
  - Change some comments
  - Recognize holes in plex structure.  Formerly this could lead to
    incorrect write to the plex.  Return REQUEST_DEGRADED on a read
    request, but carry on to the bitter end on a write request, and
    mark the requests for the inaccessible subdisks with
    XFR_BAD_SUBDISK.
  - return REQUEST_EOF if the requested transfer goes beyond the end
    of the plex.  This is not an error, since other plexes may go
    further into the volume address space.

build_read_request:
  Handle REQUEST_DEGRADED returned from bre().

sdio:
  Lock buffer before issuing the requests.

25 years agoImport RAID-5 code.
grog [Sat, 7 Aug 1999 08:11:22 +0000 (08:11 +0000)]
Import RAID-5 code.
Add Cybernet copyright.

OK'd-by:     Chuck Jacobus <chuck@cybernet.com>

25 years agoImport RAID-5 code.
grog [Sat, 7 Aug 1999 08:09:36 +0000 (08:09 +0000)]
Import RAID-5 code.
Add Cybernet copyright.

OK'd-by:     Chuck Jacobus <chuck@cybernet.com>

detchobject:  Don't update the object configuration.

replaceobject: Add preliminary code.  This is not yet complete.

vinumw.h vinumparser.c:

Add keyword 'hotspare'.

25 years agoAdd definition of new ioctl (READPOL) for setting the read policy of a
grog [Sat, 7 Aug 1999 08:08:18 +0000 (08:08 +0000)]
Add definition of new ioctl (READPOL) for setting the read policy of a
plex.  The code is not yet complete.

25 years agoRemove code to accept partitions of type 'unused'. From now on, Vinum
grog [Sat, 7 Aug 1999 08:07:05 +0000 (08:07 +0000)]
Remove code to accept partitions of type 'unused'.  From now on, Vinum
will only accept partitions of type 'vinum'.

format_config: Use the new %q format option in kvprintf, thus getting
  rid of some of the filthiest code I've written in a long time.  Also
  remove the lltoa() function.

With-great-thanks-to: peter

format_config: Accept the fact that a subdisk might not be attached to
  a plex, and save the config correctly.

vinum_scandisk: Scan all slices on a drive with a Microsoft partition
  table.  Only look at the compatibility slice if nothing was found in
  the Microsoft slices.

  This change removes a frequently employed method of shooting
  yourself in the foot: people would decide that the Vinum drives
  belonged on their own slice, and they wouldn't be able to start the
  subsystem after a reboot.  Documentation updates to follow.

25 years agoClose a memory leak.
grog [Sat, 7 Aug 1999 08:06:05 +0000 (08:06 +0000)]
Close a memory leak.

25 years agovinumopen: add code to open raw subdisks. This is now used to
grog [Sat, 7 Aug 1999 08:05:46 +0000 (08:05 +0000)]
vinumopen: add code to open raw subdisks.  This is now used to
initialize subdisks.  Probably the plex-related subdisk type will die
a death.

vinumconfig.c:

Accept (and ignore) kernel state information in userland config
files.  This saves a lot of error recovery and also makes it possible
to use the output of printconfig to create new configuration.

Remove checkdiskconfig().  It wasn't needed any more.

Start adding support for hot spare drives.  You can't put anything on
them (yet).

Change message formats from %lld to %qd.

get_empty_sd: Initialize size to -1.  Previously this was done in
config_subdisk, which is the wrong place.

start_config: set current drive, plex and volume to -1, thus stopping
  update configurations from taking their defaults from old configs.

25 years agoRework many comments.
grog [Sat, 7 Aug 1999 08:05:19 +0000 (08:05 +0000)]
Rework many comments.

requeststatus:
  add REQUEST_DEGRADED status

struct rqinfo:
  add members devmajor and devminor to compensate for lost dev_t.

25 years agoWe don't end up checking for a return value of EFAULT from the copyinstr()
green [Sat, 7 Aug 1999 05:33:35 +0000 (05:33 +0000)]
We don't end up checking for a return value of EFAULT from the copyinstr()
in the pathname translation procedure. This proves fatal, and can be
easily fixed. This or a similar change needs to be committed to svr4_util.h
and ibcs2_util.h. I will update ibcs2_util.h, if noone else thinks of a
better way to do this, in the same manner. I will leave svr4 to the
respective maintainer.

This closes the problem of the only crash I've been able to produce as
a user recently, except for (currently not-in-the-source tree) fd
table sharing fixes. Thanks goes to pho for his stress-testers.

25 years agoRe-enable play-nice-with-DHCP option now that bpf is the default
jkh [Sat, 7 Aug 1999 01:43:39 +0000 (01:43 +0000)]
Re-enable play-nice-with-DHCP option now that bpf is the default
in GENERIC.  Now a dhcp-configured system will both install and boot
initially with proper interface configuration.

25 years agoEnable bpf by default. There was no significant dissention to my proposal
jkh [Sat, 7 Aug 1999 01:42:08 +0000 (01:42 +0000)]
Enable bpf by default.  There was no significant dissention to my proposal
of 2 weeks ago that this be done, and anyone who wishes to make bpf more
selective according to securelevel or compile-time options is more
than free to do so.